[ofa-general] [PATCH] IB/ipoib: refresh paths instead of fushing them on SM change event
monis at Voltaire.COM
Thu Jun 19 08:11:25 PDT 2008
Moni Shoua wrote:
> The patch tries to solve the problem of device going down and paths being
> flushed on SM change event. The method is to mark the paths as candidates for
> refresh (with a valid flag) and wait for ARP probe to start a flow of path
> lookup that leads to patch query which ends up in path (and address handle)
> The solution requires a different and less intrusive handling of SM change
> event. For that, the second argument of the flush function changed it's meaning
> from boolean flag to a level.
> In most cases, SM failover doesn't cause LID change so traffic won't stop. In
> the rare cases of LID change, the remote host (the one that hadn't changed
> its LID) will lose connectivity until paths are refreshed. This is no worse than
> the current state. In fact, preventing the device from going down saves packets
> that otherwise would be lost.
> Signed-off-by: Moni Levy <monil at voltaire.com>
> Signed-off-by: Moni Shoua <monis at voltaire.com>
This issue is important to us.
I'd appreciate if you review it and make an opinion.
I understand it's kind of late now to make it to 2.6.26 but we hope to
have it in 2.6.27
More information about the general